Kitz Forum
Computers & Hardware => Networking => Topic started by: kitz on June 22, 2014, 12:07:26 AM
-
I recently had in my mitts on a dual band 5GHz wifi router, so thought I'd do some testing to see how it compares with my 802.11n 2.4 GHz
Thought someone may be interested in how it compares in a real world situation.
Tests were performed using speedtest.net London Namesco server on a windows 8.1 laptop & ipad2
The summary shows the mean average of 3 tests at each location. If any particular test showed an unexpected result a further test was run.
I'll attach the results summary and let you draw your own conclusion, suffice to say the only real benefit of 5Ghz was if using the laptop in the same room as the router. Other than that well..... it was totally useless in the garden. :(
-
That's interesting, because my 5ghz wifi (802.11a) works fine anywhere in the house or garden, just as well, or maybe marginally better, than my 2.4ghz g wifi (both networks powered by a HP procurve WAP).
-
If yours is ac, that may make some difference?
Unfortunately I have very little info about the other router other than "2.4GHz 300Mbps and 5GHz 300Mbps connection"
The ipad wouldnt connect to both the 2.4GHz and the 5 GHz connection, so that will have hampered things too as I had to manually select which one I wanted to connect to. This is a restriction of the ipad.
-
PS
Im not sure if this is co-incidence or not.. Normally my phone will last a couple of days without charging. But whilst the other router was in use I had to charge it slightly more often. Has anyone else seen similar or is there an explanation.. or was it just one of those things?
-
It's not 802.11ac, it's 802.11a, 54mbps 5ghz. It never really caught on.
-
My Billion 7800DXL has a choice of 2.4 GHz or 5 GHz (5 GHz by default) and I initially found that the WiFi wasn't as good as I hoped. But switching to 2.4 GHz made a big improvement in signal strength and stability. This was over a distance of about 10 metres and through two brick walls and a floor.
-
Interesting results. I think it was always accepted that 5 GHz would have poorer range, but it is meant to offer better bandwidth.
I also experimented with it using the 7800DXL, but abandoned it in favour of 2.4. That followed a test of walking up my drive, and down the garden, to identify the cut off point - 2.4 won by a good margin. I have now disabled 5GHz.
But I vaguely recollect however that I also tried testing file transfer rates, copying between the MAC Mini and the file server. Normally it's connected by cable, but I'm pretty sure I tried WiFi when I got the Billion just to compare, and got absolutely 'wow!' transfer speeds at 5GHz. I'll try and repeat that experiment later today, and share the results.
-
Thanks for the comments and your own observations.
One of the things that triggered the testing was the poor signal in the garden. I was very surprised as it was the first time I could recall not being able to get any signal at all. Even on my old st585 be box I could still use my laptop from the patio. Granted the router location has changed since then but not by much only a couple of meters to the corner of the room.
5ghz still reaches my lounge corner previous 'dead spot' ok, although I've long wondered if that's not helped by some sort of interference coming from a neighbours property.
Yes please do 7lm.. Are they in the same room?
-
IF all else is equal, 5 GHz is 6 dB worse due to frequency, maybe a dB more for worse NF and whatever the bandwidth ratio is, but set against probably a dB or two for a more efficient aerial.
But as Kitz says " If yours is ac, that may make some difference? " - As most WiFi aerials are monopoles with an earth "by guess and by God !", an AC power line could make a lot of difference in some directions.
-
I'll try and repeat that experiment later today, and share the results.
OK, here we go, crude and probably not the best tests, but…
1gig file (1024^3 bytes), passed to/fro via Samba share. Fedora server at remote end, accessed by 2009 Mac Mini , OS/X 10.8.5 client, WiFi via Billion 7800DXL.
No special effort to 'quieten' either system, update checkers and other housekeeping tasks may have been chattering away.
File copied by 'cp' command, source file 'touched' between copies. Wireless distance about 15ft, strictly through two layers of stud & plasterboard walls, but also obscured visibility thorough open door.
Gigabit lan, 3 runs each way...:
read time write time
33sec 18sec
32sec 18sec
32sec 19sec
I reckon that's average read rate circa 268Mbps, write rate circa 477Mbps
2.4GHz wireless…
160sec 138sec
154sec 160sec
156sec 138sec
I reckon that's average read rate for 2.4 GHz circa 52.5 Mbps, write rate circa 56.0 Mbps
5 GHz wireless…
130sec 125sec
136sec 145sec
151sec 145sec
I reckon that's average read rate for 5 GHz circa 59.0 Mbps, write rate circa 59.0 Mbps
Reconsidering my previous comments, it's actually the 2.4GHz rates that I find to be a 'wow', being about a five fold improvement compared to the last time I attempted to measure WiFi rates, between an old DG834GT and a 2005 Dell laptop. But the 5GHz was only marginally better.
The OS/X network utility claimed zero errors during the duration of testing, which I don't entirely trust.
Feel free to check my arithmetic, or to point out flaws in my method, I'm definitely thicker than I used to be and rarely take offence. :)
-
Thanks 7LM, like you say not that much faster with 5 GHz either, but at least some slight improvement.
Perhaps mine is duff because as soon as it goes through one wall the speeds deteriorate quite quickly.
-
IMO its difficult to get a good comparison due to the differences in the way 2.4Ghz, 5Ghz and the equipment being used works, for example
General rule of thumb the higher the frequency the lower the distance a signal will travel compared to a lower frequency but the higher the frequency the more data can be transmitted.
At 2.4Ghz reflected signals are considered interference and ignored but 5Ghz will use these signals to improve performance and provide higher data transfer at a greater distance compare to 2.4Ghz etc
If your AP has MiMo (needed to get the higher data transfer rates) and one WiFi station does and anther doesn't then this will adversely affect your results.
Don't forget that .11g equipment will connect to .11n but you wont get the throughput
You can get huge differences in through put just by changing the equipment you are using
Windows Lies; it doesn't show true values
And lastly home WiFi never perform's as well as business WiFi but then home WiFi doesn't cost as much ;)
-
I couldn't resist doing some more experiments, again using the MAC Mini and the 7800DXL.
1) With router upstairs, in furthest cornet of house (still attached by cable to the main LAN):
5GHz would not connect at all to begin with, until I experimented with a few different antenna positions. Eventually, it connected and average a rate of 5.4 Mbps, downloading a 100MByte file.
2.4GHz connected first time, and reliable connected regardless of antenna positions. But the data rate plunged to 1.2Mbps. I tried overriding the channel to various others, it made no difference.
2) With router sat on desk adjacent to the MAC:
5GHz averaged 36.32 Mbps - that's slower than it is at 15 feet away!
2.4Ghz averaged 53Mbps, similar to 15 feet away.
Thinking about it, the result from upstairs, where the flaky 5GHz was faster than the stable 2.4GHz may be explainable. The signal is weaker and prone to dropouts but, once connected, there should be less interference and so 5GHz may 'win'.
I am at a loss to explain the results right next to the router, where 5GHz was distinctly slower. ???
I am also at a loss to explain Kitz's results.
My own experiments have, I'm afraid, also rekindled my long standing scepticism about WiFi. It serves an indispensable purpose in providing web-browsing access for mobile devices. It is also capable, when conditions are right, of impressive data rates. But for any 'serious' home networking it can't be relied upon and falls short of reasonable performance. That is with my devices and in my home at least, which is a modern home with only stud & plasterboard walls and I promise is not that big. :(
-
Those are pretty low speeds kitz - was it 802.11n on 5GHz and did you enable the full channel bandwidth?
For example pretty much every 802.11n adaptor I see is configured for 20MHz b/w as default (device properties usually allows a change) rather than 40/80MHz.
I told you the real-world speeds I was getting with the Ubiquiti AP - the USB2 rate was the limiting factor - but that only happened after I enabled 80MHz channels.
As part of this I discovered that the wife's lappie was set to 20MHz b/w so a quick change in device properties & speeds nearly doubled on 2.4GHz, never mind 5GHz.
-
My own experiments have, I'm afraid, also rekindled my long standing scepticism about WiFi. It serves an indispensable purpose in providing web-browsing access for mobile devices. It is also capable, when conditions are right, of impressive data rates. But for any 'serious' home networking it can't be relied upon and falls short of reasonable performance. That is with my devices and in my home at least, which is a modern home with only stud & plasterboard walls and I promise is not that big. :(
You're wrong but that's because :
a) the manufacturers have always published PHY rates and you will only get 60-70% MAC efficiency so its been too much smoke'n'mirrors in the consumer market;
b) as mentioned above most 802.11n adaptors default to 20MHz b/w operation only
c) you're not spending enough money - I'm not kidding.
I have one of these :
http://linitx.com/product/ubiquiti-unifi-uap-ac-1300mbps-80211ac-24ghz5ghz-access-point/13806
and prior to that I thought as you did.
I don't now - I can max out the USB2 interface to a 802.11ac adaptor when the AP is in the same room. That's 280Mbps or so DATA TRANSFER. I put that in capitals because I simply couldn't believe it. I'm pretty sure that with an internal adaptor I could get 400Mbps burst speeds.
I can get (data, not PHY) speeds of over 120Mbps @ 5GHz through wooden floorboards/associated plasterboard but if you have any sort of blockwork/metal (pipes in my case) in the beampath (on any antenna) then it drops fast.
I bought the thing on the recommendation of RevK, and apart from it running surprisingly hot (vents may have helped here IMHO :P ) its been so trouble-free I am giving serious consideration to getting one of their external units. The AP software isn't that user-friendly but is comprehensive - eg you can have a rate-limited guest VLAN at the click of a button. The AP runs Linux & most of the advanced stuff assumes people would rather run a script than load up a "web interface".
tl;dr wireless works if you pay for proper RF engineering, which is the business-end of this sort of kit
-
http://www.cisco.com/c/en/us/products/collateral/wireless/aironet-3600-series/white_paper_c11-713103.html
Table 2 for those wondering about the effect of number of antennae (spatial streams) & bandwidth then this shows you what's what.
-
[/i]
2) With router sat on desk adjacent to the MAC:
5GHz averaged 36.32 Mbps - that's slower than it is at 15 feet away!
2.4Ghz averaged 53Mbps, similar to 15 feet away.
This suggests to me that your aerials have nulls at the side near the router. Try averaging over 360 degrees at say 30 degree intervals ( if you have nothing better to do !)
Seriously there is no such thing as a fully omnidirectional aerial. (If there was it would be infinitely small and have zero radiation efficiency ! )
-
[/i]
2) With router sat on desk adjacent to the MAC:
5GHz averaged 36.32 Mbps - that's slower than it is at 15 feet away!
2.4Ghz averaged 53Mbps, similar to 15 feet away.
This suggests to me that your aerials have nulls at the side near the router. Try averaging over 360 degrees at say 30 degree intervals ( if you have nothing better to do !)
Seriously there is no such thing as a fully omnidirectional aerial. (If there was it would be infinitely small and have zero radiation efficiency ! )
Given the nature of MIMO its inevitable this happens at very short ranges - unless you have really deep pockets & want to go for "steerable" antenna arrays :
http://www.ruckuswireless.com/products/zoneflex-indoor/zoneflex-7363
Even then ....
-
My own experiments have, I'm afraid, also rekindled my long standing scepticism about WiFi. It serves an indispensable purpose in providing web-browsing access for mobile devices. It is also capable, when conditions are right, of impressive data rates. But for any 'serious' home networking it can't be relied upon and falls short of reasonable performance. That is with my devices and in my home at least, which is a modern home with only stud & plasterboard walls and I promise is not that big. :(
You're wrong but that's because :
a) the manufacturers have always published PHY rates and you will only get 60-70% MAC efficiency so its been too much smoke'n'mirrors in the consumer market;
b) as mentioned above most 802.11n adaptors default to 20MHz b/w operation only
c) you're not spending enough money - I'm not kidding.
I have one of these :
http://linitx.com/product/ubiquiti-unifi-uap-ac-1300mbps-80211ac-24ghz5ghz-access-point/13806
and prior to that I thought as you did.
I don't now - I can max out the USB2 interface to a 802.11ac adaptor when the AP is in the same room. That's 280Mbps or so DATA TRANSFER. I put that in capitals because I simply couldn't believe it. I'm pretty sure that with an internal adaptor I could get 400Mbps burst speeds.
I can get (data, not PHY) speeds of over 120Mbps @ 5GHz through wooden floorboards/associated plasterboard but if you have any sort of blockwork/metal (pipes in my case) in the beampath (on any antenna) then it drops fast.
I bought the thing on the recommendation of RevK, and apart from it running surprisingly hot (vents may have helped here IMHO :P ) its been so trouble-free I am giving serious consideration to getting one of their external units. The AP software isn't that user-friendly but is comprehensive - eg you can have a rate-limited guest VLAN at the click of a button. The AP runs Linux & most of the advanced stuff assumes people would rather run a script than load up a "web interface".
tl;dr wireless works if you pay for proper RF engineering, which is the business-end of this sort of kit
The 7800DXL is configurable for 20/40 bandwidth, I had 40 enabled for my testing. I also confirmed (with OS/X utioities) 802.11n was actually in use, which it was and also noticed a '+1' alongside the channel number.
It may of course be my Mac Mini that's the bottleneck, it's getting on a bit. But that actually raises another point... you can invest big bucks in the access point, and possible get a much better AP. But the connected devices also have their own RF front ends and, say, a tablet or laptop, or a TV with built in WiFi, you're stuck with whatever RF components the manufacturer provided. Which is probably cheap and basic, and I argue that may deny you much of the gains from the super-duper AP.
Don't get me wrong, I couldn't live without WiFi. I'm using it right now, sending this reply from an iPad. And I'd spend whatever was needed to make it work. But for myself, all I need from WiFi is web browsing. I'll stick with gigabit ethernet and (cheap) Cat 5 for anything that's bandwidth hungry, such as my MythTV based media, which needs to be able to reliably sustain Full HD data rates between server & frontend :)
-
Don't get me wrong, I prefer wired :)
It was just one of the few times I've been genuinely gobsmacked with a new bit of kit - a sort of "oh THAT'S how it should work" moment :o
In the overall scheme of things I'd say its cheap for what it delivers - can push out 3-4 concurrent HD streams (mixed 2.4/5GHz 802.11n/ac) here with no stuttering in normal use.
You're quite right about clients but if the AP is the bottleneck then all the new shiny stuff seems slow anyway. The shiny stuff usually costs more than the Ubiquiti AP I linked anyway :D
-
WiFi always strikes me as similar to "Cordless" - a nice sounding effort saving idea, till you meet the need to charge batteries, replace them or problems of possible explosion and toxic waste.
Yes they both work but not with zero effort !
Just had a look at the "Help with WiFi" instructions for my Router. In effect "If you have speed problems try Ethernet "
Pragmatic advice for most people; I might have added " or read Hertz's 1888 Thesis !"
-
Semi-on-topic....
I've been playing with some (mainly Samsung) Android devices lately & unless you mess about editing hex in a config file then they are very reluctant to use the 5GHz band.
From what I can tell if the SSID is the same for both 2.4GHz & 5GHz radios then the Android device will only use the 5GHz band if the signal strength is a few dB greater than the 2.4GHz signal. This is pretty nonsensical given the same signal strength would usually result in a less congested RF environment at 5GHz - and that in turn would result in faster speeds. Edit - and of course the signal strength of both is much the same in a single room so where there's a perfect signal it still uses the 2.4GHz band ???
So the only way to do this on a consumer-grade access point is to have 2 SSIDs - one for 2.4GHz and one for 5GHz but that's not great really. It works but just seems like a bodge.
The AP I have can recognise clients & force them to join a "group" and as part of that group you can specify which radio they should be using & when to switch - which seems to work so far.
Seems rather dumb to me that Android (not just Samsung) devices seem to default to this behaviour. Edit2 - this is on both KitKat & Jelly Bean builds.
-
Seems rather dumb to me that Android (not just Samsung) devices seem to default to this behaviour. Edit2 - this is on both KitKat & Jelly Bean builds.
Sorry I should already know how, but if you tell me how to tell which signal the phone has connected to, I'd be glad to try that here and see if I get same results.
Even more off-topic, I have already concluded that Samsung's Android offerings are data-flawed, in relation to 2g/3g (I did say Off Topic). Mine works brilliantly on 3g but that's not always available, and Samsung seem to overlooked the fact that we sometimes visit the countryside and so 2g data needs to work too. In GPRS-only locations where many other phones, from a ten year old Motorola flip to a modern iPhone all work perfectly, the Samsung just sits there, flashing its activity indicators for an eternity, but hardly ever actually completes the task.
-
I have no idea how you tell. If there is anything which does indicate what radio you're connected to then I'm buggered if I can find it :)
The only way I can tell is from the access point which tells me which clients are connected to each radio.
Edit - oh & I have an ancient INQ phone which isn't remotely "smart" but does keep working for several days without charge so can't help with the Samsung phone issue. Daughter has one but try taking "her precious" away to play with :D
-
I already thought of another way. The Billion supports multiple virtual lans, each of which has its own subnet and DHCP pool. If I put the two WiFi bands on different vlans, but with same SSID, I can then tell which one it has conneted from the assigned DHCP address :graduate:
Or I could do it your way, which sounds less bother :D
But it's going to have to wait probably tomorrow, will report back. I'll be trying my iDevices too as it's an interesting excercise.
-
The problem for me is that I can't really use multiple SSIDs as I want to use zero handoff* which requires a single SSID (and channel) across multiple APs. That way you don't end up with time-critical stuff like audio/video stuttering (or failing) when you roam around - I'm thinking basically inside/outside scenario where you will lose 5GHz connectivity fast unless there's an outside AP. The future seems to be more wireless devices whether I like it or not ;)
Fortunately I can force clients to use a specific radio from the APs but I'm not 100% convinced that won't bugger up zero handoff in certain scenarios on the 5GHz band. I suspect I'm going to end up editing some of the files in /system/etc/wifi on the Android kit eventually, which isn't exactly an elegant solution.
* http://community.ubnt.com/t5/UniFi-Frequently-Asked-Questions/UniFi-What-is-Zero-Handoff/ta-p/412719 & a "frendlier version" http://www.lansor-it-services.com/wi-fi/ubiquiti-zero-handoff-seamless-roaming/ Somewhat amazingly (to me anyway) it actually works :)
-
Well mine was a useless endeavour, my Samsung (Fame) doesn't even see the 5GHz network, it doesn't seem to be supported! :blush:
Same for 4th generation iPod touch, 5Gz apparently not implemented.
iPad 2 and latest iPad mini, both work on 2.4GHz as well as 5GHz, and generally seem to have a predilection for 2.4, given the choice. It is vague though, at least once the iPad 2 chose 5GHz when 2.4 was available too, not sure what was different. I wonder if recent history of connections may play a part in choice of networks?
The Mac Mini actually seems to prefer 5GHz when it has a free choice, but doesn't seem to cope too well when I disable the connected signal, one on occasion it went into a strop that lasted several minutes, timing out on every attempt to register. Then it suddenly worked again, after fiddling with various things that shouldn't have had any effect.
-
What are you blushing for? My mobile doesn't have wireless at all :D
-
I am not that surprised by kitz results.
Few thigns to consider.
First how good the wireless devices are. e.g. if a 2.4ghz device can only do max 72mbps anyway due to the hardware, and 2.4ghz is running well, then 5ghz may well be slower.
My galaxy s3 can currently comfortably run faster on 5ghz than 2.4ghz I say 'currently' because on a different firmware it ran like rubbish due to some kind of bug in the android 4.1.x on that phone. Since 5ghz is less popular that is more prone to bugs/software issues. The s3 has a 80mhz band capable reciever. It forces itself to 40mhz only mode on 2.4ghz even if the router is configured to allow 80.
My AC dongle which I got from BT when I was trialling for them, is a no brainer, as it will clearly be way faster on 5ghz than 2.4.
I no longer have 5ghz to myself now as well, a VM customer is using it now. But at least isnt super crowded yet like the 2.4 round here. However I fear with isps offering dual band routers by default now this situation wont last its only a matter of time and 5ghz has less useable channels than 2.4. Since alot arent selectable.
Somethign else i discovered.
after I invested in a AC router. It became apparent so many mainstream devices lag behind on tech. XBOX360 2.4ghz only, wii-u 2.4ghz only, ps4 2.4ghz only and thats a late 2013 kit, set top boxes, 2.4ghz only. Also most of these devices listed are 40mhz wide max so 72mbps max on N.
Apart from cutting edge android phones and wifi dongles what has AC?
-
Just found this a bit late whilst looking for something else.
I, too, have a 7800DXL and just found an annoying fault: 2.4 band wifi only uses channels 5 to 13 (the manual, equally daftly, says 1-7 only).
Have I missed something or is this an utterly silly problem?
-
@Mardler
I think that'll be due to the arrangement of 40MHz of bandwidth in the 2.4 GHz band, and whether what the manual refers to as the "sideband" (second 20MHz portion) is above or below the main 20Mhz channel.
----
The s3 has a 80mhz band capable reciever. It forces itself to 40mhz only mode on 2.4ghz even if the router is configured to allow 80.
Yes I know it's a very old post, but it's full of things that are horribly wrong. I'm pretty sure 2.4GHz only has 11n which only has 20 or 40 MHz channel widths.
Also most of these devices listed are 40mhz wide max so 72mbps max on N.
It's not just the bandwidth, it's the number of spatial streams. 11n is:
1 stream, 20MHz is 65, or 72 Mbps with SGI (Short Guard Interval).
1 stream, 40MHz is 135, or 150 Mbps with SGI.
-
yeah I meant its capable of 40 and locks to 20, sorry. :)
the s3 has 2 spatial streams, as does the s5 and s7. My LG G5 only had one stream, and my oneplusone phone also only has one.
-
Thanks, guys but I answered it myself last night:-
Switch from 20/40 to 20. Channels now 1-9.
Select 1 and Apply.
Switch from 20 back to 20/40.
Easy but why isn't this in the manual?